of adsorption rise with the surface coverage slowly and a1)l)roxiinately linearly. This prrrnits t,he reliable ext,rapolation of these values to zero coverage ('7). I3ut', in the case of alcohols such ext,rapolation is impossible, as a t all coverages, available a t the calorimetric measurements a t room temperature, the heats of adsorption of alcohols already include the energy of mutual hydrogen bonds (1,f I ) . I n accordance with this, as we can see from the left part of Figure 3, in the case of hydrocarbons, the results of calorimetric and isosteric determinations of heats of adsorption coincide with the gas chromatographic data. 13esides, all these experimentally obtained values are close to the calculated potential energies of nonspecific adsorption of isolated molecules which we obtained by the theory of molecular int,eractions a t short dist'ances (1, 22). In the case of alcohols, we have received quite a different picture. The heats of adsorption, determined gas chromatographically, are 4.8 Kcalories per mole less than the heats of adsorption, determined calorimetrically a t room temperature, but are close t'o the theoretically calculated values of potential energy of nonspecific adsorption of isolated molecules of these alcohols on graphite. On the other hand, we can notice that the heats of adsorption of isolated molecules of alcohols determined by t,he gas chromatographic method are much lower
t'han t'he heats of condensation because t,he latter include the energy of specific interaction (mutual hydrogen bonds between the alcohol molecules in a liquid state). Such great differences in the heats of adsorption determinations of alcohols obtained by different methods, exist because the gas chromat,ographic method involves very low surface coverages. This means that the molecules of alcohols are not' associated and are adsorbed on the surface of carbon black as isolated molecule3. The difference between t,he calorimetric values a t room temperature which already include the energy of association, and the gas chromatographic values is equal to t'he energy of association of alcohols on the surface of t'his adsorbent. We can see that, in this case, t'his energy is equal to about 5 Kcalories per mole which is most'ly due to the energy of mut,ual hydrogen bonds between the alcohol molecules. Thus, the gas chromatographic method for determination of heats of adsorption on graphitized carbon black permits one to determine the energy of adsorption of isolated molecules even if they are capable of mutual association. I n combination with t.he determination of heat of adsorption a t higher coverages (for example, with calorimetric methods a t room temperature) the gas chromatographic method permit's one t'o determine also the energy of association in the adsorbed layers on nonpolar surfaces.
LITERATURE CITED
(1) Avgul, S . S . , Kiselev, A . Lygina, I. A . , Izv. A c a d . .Yauk I'SSIZ, Otd. K h i m . AYazLk,1404 (1961). ( 2 ) Rarrer, R. AI., Ries, L. Y. C., l'rans. Faraday Soc. 57,999 (1961). ( 3 ) Belyakova, L. I)., Kiselev, A . V., Kovaleva, S . V., Doklady A k a d . S a u k V S S R , in print (1964). (4) Brunaur, S.> Emmett, P. H., Teller, E., J . Am. Chern. SOC.60, 309 (1938). (5) Gale, R. L., Beebe, R. *4., J . Phys. Chern. 68, 555 (1964). (6) Halasz, I., Horvath, C., .\-atlire 197, 71 (1963). (7) Isirikyan, A . A , , Iiiselev, 4. \-,, J . Phys. Cherri. 65, 601 (1961). (8) Kiselev, A. \-., Schtscherbakova, K. I)., Abhandlungen der Ileutschen Akademie der Wissenschaften zu Berlin, Gas-ChromatograIhie 1961, Akademie\.erlag, Berlin (1962), S. 207. (9) Kiselev, A . (:as-Chromatographie 1962, ed. by AI. van Snaay, p. XXXI\-, Butterworths, London, 1962. (10) Kiselev, A . \-., Paskonova, E. A , , Petrova, R. S., Schtscherbakova, K. I)., J . Phys. Chem. (.140scow), 38, 161 (1964). (11) Kiselev, A . IT., Quart. Rev. 15, 99 \7,j
\7.j
i1861 - ). , \ - -
(12) Kiselev, A. Y., R e z . Gen. Caoutchouc 41, 377 (1964). (13) Littlewood, A . E., Phillips, C. S. G., Price. 1). T.. J . C'hein. Soc., 1480 (1954). (14) Ross, S., Saelens, J. K., Olivier, J. P., J . Phys. Chevn. 66,696 (1962).
RECEIVEDfor review April 13, 1964. Accepted hpril 23j 1964. 2nd h t e r national Symposium on Advances in Gas Chromatography, University of Houston, Houston, Texas, lIarch 23-26, 1964.
Automatic Capillary Gas Chromatography and Sampling of Distillation Products D. W. GRANT The Coal Tar Research Association, Oxford Road, Gomersal, Leeds, England
b An automatic sampling valve has been developed which, when inserted a t a single point into a gas or vapor system, extracts small repeatable samples and transfers them to a capillary column gas chromatograph for analysis. Details of the valve and associated equipment are given together with results of laboratory tests with b en Zen e, to I uen e, x y Ien es, pyridine, picoline, phenol, and cresols. Minor impurities in these products can b e continuously determined during distillation and this is considered to b e the main potential application of the apparatus. The valve has been successfully operated a t temperatures of up to 200" C. and results are given for a vapor system at subatmospheric pressures. The limitations of the valve in the latter connection are demonstrated.
TYPE of sampling valve is described, together with a compact form of capillary gas chromatograph which is particularly applicable to the problem of sampling distillation or static systems. NEW
EXPERIMENTAL
Apparatus. T h e essentials of t h e system are shown in diagrammatic form in Figure 1. T h e sampling valve is operated electromagnetically by pulses from a time switch (Sangamo 277, form 1) which can be adjusted to operate u p to 12 times per hour. Pure argon is used as the carrier gas and is drawn from a cylinder a t about 20 p.s.i. This is dried and the pressure is further reduced by a diaphragm pressure controller to the column operating pressure, (5 to 10 p s i ) . - i n argon
purge and stream splitting system are employed in the usual way for capillary column operation. The stream splitt'er also ensures that t,he sample is carried rapidly from the sampling valve to the column. I n practice the amplifier, column unit, and gas controls, etc., are contained in a separate cabinet, which can be isolated from the sampling valve. Gas connections between the cabinet and valve are made by 1.5-mm. i.d. polytetrafluoroethylene (P.T.F.E.) tubing of any reasonable length provided a sufficiently fast by-pass gas rat'e is used to minimize diffusional effects. Condensation of sample is avoided in the return line by a low voltage heating wire which passes through the center of the P.T.F.E. tube. COLUMN UNIT. The column unit was designed for capillary column operation in a robust yet portable form in view of its possible use as a plant instrument. VOL. 36, NO. 8, JULY 1964
1519
__7
!
ELECTROMETER AWLlFiER
ImV HIGH SPEED RECORDER
ARGON INLET
-
A R G O N STREAM GLC APPARATUS
t b
u
RETURN SPRING
INTEGRATOR
Figure 1.
Flow diagram of automatic sampling system
Brass const,ruction is used throughout, and cartridge type insert heaters are employed. 'The column temperature is controlled by means of a cartridge bimetal thermostat and transistorized control relay. The microargon detector is sit,uated concentrically inside the thermostated chamber and contains a 200-me. tritium source. Sufficient space iq allowed for about 100 feet of capillary column and this is mounted on a stirrup arrangement. Glass capillaries ( 1 ) are employed and are connected t o the inlet port and detector by means of P.T.F.E. tubing. l'he unit separates into two parts on the guide rails to allow access to the column and detector. l'he same movement also disconnects the high voltage supply to the detector. Provision is made for the manual injection of samples via hypodermic syringe for the purpose of calibration, etc. 'The detector is operated a t voltages of up to 1200 volts, and t8heionization current is amplified by an electrometer amplifier of conventional design with a maximum input impedance of l o 9 ohms and niininium output attenuation of 230:l. A 1-mv. high speed recorder (Honeywcll 13rown 1 1 , 4 second) is used to trace the chromatograms. This is fitted with a modification which allows a ret,ransniitted voltage t o be fed to a separate integrator should this be desired. +\lvToMATIC SAhlPLTlVG V.ZLVE. Camponent parts of the sampling valve include the sampling head and piston, both of stainless steel const,ruction, and the sliding surfaces honed to a maximum clearance of 0.001 inch. h miniature solenoid is employed operating at, 250 volts a.c. and ha\-ing a movement' of inch (Electromethods Type 170). This is mounted on a brass plate attached to the valve. The instrument is situated in the vapor system and spaled in position by means of the standard 1334 cone shown. This is suitsable for laboratory testing purposeq provided a P.T.F.E. sleeve is employed to rpduce the possibility of fracture with glass ai)paratus. For plant operation a flanged joint would be used. 'Two lorn-voltagc 30-watt cartridge heaters a w insertcd vertically in the wall of the t-alve to prevrnt cwndensation i n thc case of high h i l i n g conil,ountls. 'l'hest allow the temperature to be raised t o about 200' C . 1520
ANALYTICAL CHEMISTRY
H O N E D PISTON STAINLESS STEEL SAMPLING H E A D (Immersed in vapo to be sampled) YON-RETURN
Figure 2.
Glass capillaries were used in the present st,udies and these were cleaned thoroughly with dry ether before use. The capillaries were then coated by filling loo/, of their length with a 57, s o h tion of the stationary p h a x in ether and then blowing at 5 to 10 cm.;second through their length with dry nitrogen. The blowing was continued fur 1 to 2 hours to remove the solvent. Mode of Operation. AX diagrammatic form of the valve is shown in Figure 2. I n the unenergized position, argon flows through the valve via the groove which is machined in the piston and proceeds to the chromatograph. At t,he moment the solenoid becomes energized the piston commences to travel upward and immediately seals off the argon flow. .it the same time some vapor is drawn through the nonreturn valve into the sampling head by the vacuum created by the movement of the piston. As soon as the lower edge of the piston reaches the level of t.he argon inlet port the pressure is re-established immediately sealing the ball valve and trapping a known volume of vapor inside the sampling head. This is immediately conveyed to the chromat'ograph by the argon flow for analysis. When the solenoid is de-energized the pist,on returns t,o its normal position under the action of the return spring. coating of molybdenum disulfide on the two sliding surfaces aids the smooth operation of the valve and prevents binding effects. The stream splitter in the column unit is adjusted to allow a by-pass rate of 50 to 100 ml.;'minute tmoensure a rapid transfer of the sample from the valve to the capillary column. The total sampling time is thereby minimized to a fraction of a second. In practice the solenoid is allowed to remain operating for 2 to 3 seconds to avoid any possibility of incomplete sampling. 'The theoretical volume of sample taken b!. the \.alve is 0.4 ml. but the
Cross section of sampling valve
actual volume is somewhat less than this because some of the sample escapes before the one-way ball valve reseals. The successful action of the sampling valve depends on the existence of a positive pressure differential b e t w e n the argon gas streani and the vapor system. Thus, sampling cannot be achieved if the pressure of the system is higher than the argon pressure. This, however, is an unlikely situation in fractional distillat,ion since atmospheric or subatmospheric pressures are normally used. RESULTS
Repeatability of Sampling. The equipment was tested to determine the combined repeatability of the sampling valve, stream splitting system, and final capillary column analysis. Samples of vapor were taken from a static system comprising a mixture of 40' to 60' petroleum spirit, benzene, and toluene in a 1000-ml. flask maintained a t 20' C. The conditions were : 5 minutes rm. of Hg Sampling interval: 60 Argon pressure: System pressure: atmospheric Column: 3o-foot X 0.01-inch i.d. glass capillary coated with silicone oil M.S. 550 Column temperature : 20° c. 100 ml./minute By-pass rate: Detector voltage: 900 volts 2500: 1 Attenuation :
The analyses were completed within one minute and a complete separation was achieved between benzene, toluene, and nonaromatic peaks. The toluene peak was found to be the most suitable for study purposes, and the peak height was measured for eight successive samples. The peak width was the same in each case. The results are given in Table I
together with the standard deviation value. Effect of Argon Pressure a t Constant System Pressure on Sample Size. Samples of the above mixture were taken by the valve a t different argon pressures. I n each case the pressure in the vapor system remained a t atmospheric level to ensure t h a t the mass of sample taken was a constant fraction of it,s volume. T h e toluene peak height was taken as a rnewur? of the mass of vapor sampled. Examination of the results given in Table I1 indicates that for argon pressures above about 100 mm., large variations in pressure have a comparatively small effect on the volume of sample taken. The smaller samples taken for pressures below 100 mm. are probably caused by the slower sealing of the ball valve immediately after sanil)ling. Effect of System Pressure a t Constant Argon Pressure on Sample Size. T h e effert of reduced on the effectiveness of sampling was tested in apparatus similar to t h a t previously used but equipped; with a vacuum p u m p and Cartesian manostat. Pure mesit'ylene was employed as the test c o y p o u n d and the chromatographic separation was carried out on a 32-foot capillary, coated with di-isodecyl phthalate and operated a t 95" C. with an argon pressure of 38 em. of Hg, The samples of mesitylene were eluted by the column within one minute, and samples were taken automatically every five minutes. Off-scale peaks were calculated by simple proportion from a minor impurity. The sample size decreased with decreasing pressure to zero a t a pressure corresponding t o 200 mm. below atmospheric. The failure of the valve to sample successfully a t pressures below this value can be explained by the low pressure differential which is created across the nonreturn valve on the upward movement of the piston. This is insufficient to overcome the weight of
Table I. Run S o . Toluene peak height(cn1.)
Repeatability of Automatic G.L.C. System 2 3 4 5 6 7
17.0 16.8 17.1 16.9 16.8 17 4 Average peak height = 17.0cm. Standard deviation = 0.329 cni. % Standard deviation of average = 1.94
17.1
8
17.1
the ball and hence the valve remains approsimately 10%. The efficiency of closed. the distillation was then deliberately In its present form, therefore, the lowered by increasing the take-off rate valve would be unsuitable for sampling and t'his lowered the pyridine consystems under vacuum conditions below centration still further to about 1%. the critical point of 200 mm. of Hg. and Lower conccntrations than this could some modification for this type of apbe obtained in the test, but from the plication may be necessary. chromatograms it w-as evident that Application of Automatic Valve to levels down to O.ly0 of pyridine in the Monitoring of Distillation a-picoline would be detected without Columns. AROMATICHYDROCARBONS. difficulty. The application of the equipment to The critical effect of take-off on the the monitoring of a laboratory distillaefficiency of the distillation of pyridine tion of aromatic hydrocarbons was and a-picoline was shown when take-off tested for a continuous running period was commenced at approximately of seven hours;. Results for three dis2 to 3 ml./minute. The concentration tillation systems are given in Table of a-picoline in pyridine increased rap111. idly. On restoring total reflux conITnderthe conditions of test, impurity ditions, however, the a-picoline level levels as low as 0.017~in the main returned to its original value of product could be detected. This test 0.17'. This experiment illustrates clearly demonstrated several of t'he another possible use of the sampling advantages of the equipment. Thus valve viz., as an instrument for estabthe best cut points could be determined lishing optimum efficiency conditions in directly and hence over-contamination a still, or for studying the effect of with impurities can be a v o i d d . Also, variables on fractional distillation efany variation in the still conditions was ficiencies. immediately reflected in the impurity level. Thus it was found that an increase in the boil-up or take-off rate reTable II. Effect of Argon Pressure on sulted immediately in a rapid increase Sample Size in the height of peaks caused by impurities. Argon pressure Peak height of (mm. of Hg) toluene (cm.) In order to detect trace concentrations of pyridine in a-picoline the distillation 15 1 0 38 4.5 was allowed to proceed until most of the 103 6.5 pyridine had been removed. The 223 6.4 automat,ic sampling valve was operated 345 7.9 during this time and the pyridine 465 7.0 decreased to a constant level of
Table 111.
Distillation conditions 15-inrh x 3/4-inch column packed with '/*-inch Dixon gauze rings and fitted with McIntyre still head. Sampling valve positioned in still head approximately 0.5-inch above liquid level in weir. Distillation effiriency approximately 7 theoretical plates.
1
System 1. Benzene Toluene Xylene 2. Phenol Cresols 3. Pyridine a-Picoline
Monitoring of Distillation Columns Minimum concentration Composition of of minor still head vapor component Valve (Major cnmdetected temp. ponent first) % v./v. C. Benzene, toluene 0.01 I.00 Toluene. benzene 0 01 O
Chromatographic conditions of 32-foot X 0.01-inch glass capillary Stationary phase Di-isodecylphthalate
Temp. 50
Argon pressure 20-cm. of Hg
c.
Toluene, xylene
0 1
o-Cresol, phenol
0 5
200
Di-isodecylphthalate
95
40-cm. of Hg
Pyridine, a-picoline a-Picoline, pyridine
0 1
100
Silicone oil M.S. 550
50
20-cm. of Hg
Xylene, toluene
0 1
11
VOL. 3 6 , NO. 8, JULY 1964
1521
The operation of the sampling valve has been shown to be unsatisfactory at Ilressures lower than 200 mm. below atmospheric. Phenols are normally distilled a t lower pressures than this and hence the valve could not be tested under plant distillation conditions. However, the tests shown in Table I11 were carried out with a sample of cresols containing a trace of phenol which was reflused under atmospheric pressure.
The impurity concentrationq were determined by prior calibration of the column using test mixtures of known composition added through the manual injection port by means of a Hamilton microsyringe. Samples from the weir were then chromatographed to find the actual impurity concentration and hence to establish the relationship between the peak height due to the impurity from automatic sampling and its concentration in the weir condensate.
LITERATURE CITED
( 1 ) Desty, D. H., Goldup, A., Whyman, B. H. F., J . I n s t . Petrol. 45, 287 (1959).
RECEIVED for review January 13, 1964. Accepted April 13, 1964. Thanks are due to the Director of the Coal Tar Research Association for permission to publish this paper. An application for an American patent on the sampling valve is being considered. Presented a t 2nd International Symposium on Advances in Gas Chromatography, Tniversity of Houston, Houston, Texas, March 23-26, 1964.
Use of Glass Capillary Columns with Modified Internal Area in Gas Chromatography F.
A. BRUNER and G. P. CARTONI'
lstituto di Chimica Analitica, Universitci di Napoli, Napoli, ltalia
b Glass capillary columns, after treatment with an alkaline solution, are used as adsorption columns for gas solid chromatography. They show high rerolving power and are useful for the analysis of low boiling hydrocarbons at room temperature. By coating these capillaries with different amounts of liquid phase, modified gas liquid chromatographic columns are obtained. The large internal area of these capillaries allows thin film ihickness and consequently good resolution with high liquid phase to gas phase capacity. Plots of plate height v s . linear gas velocity are reported and interpreted.
G
coated with different liquid phases have been used extensively as partition columns in gas chromatography. 13y nodifying the internal surface, they way be used also as adsorption colun-ns. Mohnke and Saffert (8) emplcyed glass capillaries, a here an internal layer of corroded silica behaves as an adsorption medium, for the separation of hydrogen isotopes at low temperature. A modification of their adsorption properties by addition of a small amount of liquid phase for the separation of polar compounds has also been reported (7)' Adsorption capillary columns were also made from aluminum tubing containing an inner surface of active alurrinum oxide (fO), or from metal tubings coated with graphited carbon black (4) or active silica ( 1 4 ) . Hal& LASS CAPILLARIES
and Horl ath ( 5 ) described cal)illnrv columns modified by coating thcii inner surface with a thin layer support impregnated with liquid phase. In this work a large internal qurface has been obtained in the glass capillaries by chemical corrosion of the silica; their behavior has been investigated as such for gas solid chromatography, and uhen coated with a liquid phase, for partition chromatography. The influence of the amount of the stationary phase in the efficiency of the columns has been studied going from gas solid to modified gas liquid chromatography. EXPERIMENTAL
Preparation of Column. Glaw capillaries were prepared from qoft glass tubing according to the method described by Desty ( I ) . The column is immersed in a mater bath, with one end connected to a reservoir containing a 20% solution of NaOH in water, and completely filled.
Table 1.
L K
=
1522
ANALYTICAL CHEMISTRY
Column Characteristics and Experimental Data
38 meters; r = 0.22 mm.: V c = 5.77 ml., S
sq. em.; __ KLI,,",
Column number Liquid phase o/c v./v. 2L', mg. d,, microns V A (n-heptane), ~ ml. (temp. = 0" C . ; p = 1atm.) V a d a (n-heptane), ml. (temp. = 0" C., p = 1 atm ) Adsorption, "/b V L I V Gx lo3 R, sq. cm./second C x 102, seconds k '(n-heutane) HETP,,, = 2 d / R C , m m T,,,, = d/R/C,cni./serond \
Present address, Istituto di Chimica Generale c Inorganica, Universiti di Roma, CittR Cniversitaria, Roma, Italia
2198 X
=
L
A slow flow rate (10 ml. per hour) of this solution is maintained during the heating (six hours at 100" C.) and during the cooling, by applying pressure (2 kg. per sq. cm.). The capillary is washed with water until no alkaline reartion is observed in the effluent, then with absolute ethanol-ether, and dried in a flow of nitrogen. After this treatment a uniform white layer is observed along all the inner surface. For coating the column with different amounts of liquid phase, solutions of squalane in ether from 1 to 10% v./v. were used, to which a small amount of a dye was added to follow the filling operation. About 5 meters of column was filled with the solution, and by applying slight pressure, it is run through a t a constant speed of about I O rm. per minute. To obtain comparable results, different amounts of liquid phase were introduced into the same column. Apparatus and Materials. All measurements were carried out in a home made chromatograph and
=
=
3750 sq. cm.
0.5; temp. = 40" C. stationary phase, squalane
2
0 01
3 2 11 2 0 04
4 35 1 0 12
5 6 55 6 0 18
14 72
14 57
17 31
28 22
35 54
14 72 100 0 0 1 0.52 3 24 0 35 4 4
12 68 87 0 68 0 os 0 55 2 89 0 42 3 8
10 57 61 2 44 0 15 1 20 3 30 0 85 3 6
7 09 25 7 65 0 15 1 25 5 80 0 8s
2 13
1 0
0 0
1 3 14
4
3 3
6 10 9 0 18 1 25 8 01 0 9S 3 i